Crypto Time-Series Momentum research pack

An aggressive-tier strategy that holds, once a week, the coins among the 20 most-traded USDT perpetuals whose 1- and 4-week trends are up, weighted by inverse volatility at 1x gross in total, and otherwise stays in cash. The numbers that mean luck cannot be ruled out (a deflated Sharpe ratio below 0.95) and its large lag behind buy-and-hold Bitcoin in 2023–2026 are disclosed in full below. It has not been verified in live trading; its AI league paper record is in progress. The pack holds the research report, the backtest source and the strategy pack for the free AvalonQuant app.

Price

99 USDT. Paid in USDT on BNB Smart Chain (BEP-20) or TRON (TRC-20). Sign in, start an order, send the exact amount shown, and the download unlocks on this page once the payment confirms.

Tier: Aggressive

Carries higher risk, though this strategy runs without leverage. Luck has not been ruled out, and the AI league paper record is still in progress.

Risk disclosure

Deflated Sharpe ratio (DSR): 0.615
This strategy was selected from many backtested candidates. Its deflated Sharpe ratio is below the 0.95 confidence level, so the backtest result may be luck and cannot be ruled out as such.
2020 Feb-Mar bear market (2020-02-01 to 2020-03-31): strategy max drawdown -18.34%, Bitcoin (BTC) -54.00%
2022 bear market (2022-01-01 to 2022-12-31): strategy max drawdown -23.95%, Bitcoin (BTC) -66.94%
Leverage: declared leverage 1x, perpetual futures (max gross leverage 1x, long only)
Max drawdown: -29.23%
Holdout window (2022-12-31 to 2026-10-05): strategy CAGR +15.29%, risk-adjusted 0.52; buy-and-hold Bitcoin (BTC) CAGR +54.87%, risk-adjusted 1.04
In this holdout window it trailed buy-and-hold Bitcoin (BTC) on a risk-adjusted basis.
AI league paper record: in progress — not a live-trading verification.

Backtest you can reproduce

CAGR	+29.07%
Max drawdown	-29.23%
Sharpe	0.93
Total return	+493.9%
Period	2019-10-12 – 2026-10-05
Universe	Top-20 USDT perpetuals by trailing 30-day volume (point in time), long only 1x
Re-run	2026-10-06 · crypto-tsmom-voltarget-research-pack-0.1.1.zip

These numbers come from running the backtest source shipped in this pack, unchanged. The window runs from 2019-10 on Binance USDT-M perpetuals to the day it runs, with a 0.10% cost on every traded leg and real funding included (at 0.20% it compounds 26.60% a year). A later re-run covers a later window, so its numbers move.

How the strategy works (sample)

On completed daily closes, executed at the next day's open: the universe is the 20 USDT-margined perpetuals with the highest trailing 30-day dollar volume (in the backtest chosen on each date from every perpetual ever listed, delisted ones included). Once a week it scores each coin by two trend votes — its 7-day and 28-day returns, +1 if positive and −1 if not — and holds the coins with a positive average score, long only, weighted by inverse 30-day volatility, at most 1x gross in total. Otherwise it stays in cash. No leverage, no shorts.

What is inside

File	What it is	Size
START-HERE.md	Start here: get the app, set it up, reproduce the backtest	2.6 KB
crypto-tsmom-voltarget-0.1.1.zip	Strategy pack for the AvalonQuant app	31.2 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/README.md	Folder overview	1.0 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/SOURCE.json	Source record	1.3 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/_shared/backtest_engine.py	Backtest engine (fills, costs, equity)	42.6 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/_shared/crypto_perp_data.py		7.7 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/_shared/perp_book.py		2.6 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/_shared/point_in_time_universe.py		6.4 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/assumptions-and-risks.md	Assumptions and risks	1.4 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/backtest/backtest.py	Backtest source (downloads its own data)	5.0 KB
research/crypto_tsmom_liu_tsyvinski_voltarget/parameters.json	Full parameter set	156 B
research/crypto_tsmom_liu_tsyvinski_voltarget/research-report.md	Research report, with the measured risk disclosures	2.9 KB

The report, the full parameter set (7 values), the backtest source and the strategy pack for the free AvalonQuant app. The app itself stays free and is provided as is.

Risks (from the pack)

Provided as is, without warranty. Not investment advice. You decide whether and how much to trade; losses are yours.
Aggressive tier. It passed Avalon's methodology checks but not every luck check. Deflated Sharpe 0.615 (< 0.95): the backtest result may be luck.
Recent underperformance. From 2023 to 2026 it trailed simply holding Bitcoin, both in return (15.29% vs 54.87% a year) and risk-adjusted (0.52 vs 1.04).
Drawdowns. Backtest maximum drawdown −29.23%; −23.95% in 2022, −18.34% in Feb–Mar 2020.
No live verification. The AI paper league record is in progress; nothing here has been verified with real money.
Costs and funding. The backtest assumes 0.10% per traded leg (fee + spread + slippage) and pays real historical funding. Higher costs (the backtest at 0.20% earned 26.60% a year) or different funding reduce returns.
Exchange risk. Perpetual futures can be delisted, halted or liquidated; this pack trades long only at 1x on isolated margin, which limits but does not remove liquidation risk.
Universe. Live, the pack ranks a fixed list of 30 liquid perpetuals by its own trailing dollar volume; the backtest ranked every perpetual ever listed. A coin outside the list cannot be picked live.
Paper first. The app runs every pack in paper mode until you switch it to live yourself.

Compare with other strategy backtests · Free Strategy Kit

Disclosure. Backtests are hypothetical simulations on historical data. They do not include every real-world cost, are not live results, and do not guarantee future returns. Educational research only, not investment advice.